agent-controller AGENTS.md

Repository instructions for Agent Controller, a system that defines AI agents in YAML and runs them through different agent runtimes using one shared interface.

In plain words
What is it for?
Use them when developing or validating the Agent Controller command-line tool, runtime adapters, agent definitions, registries, or its build and test process.
Why use it?
They explain the repository structure, setup, build, and tests so changes follow the project's supported workflow.

Instructions file for CodexOpenCode

Install

Getting it into your agent

One page per mod, every tool's command on it. A separate URL per tool would split the same page into five that compete with each other.

agentmods
npx agentmods add instructions/ccdevelopforfun/agent-controller/agents-md
Clone the repo
git clone --depth 1 https://github.com/CCDevelopForFun/agent-controller

Made for: Codex, OpenCode.

Per session 1,981 This file is loaded in full into every session.
When invoked 1,981 The same file — it is already loaded in full.
Security scan A 0 findings. Scan, not verified.
Origin original No closer match found in the catalogue.
Token cost

What it costs to keep this loaded

Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.

ModelPer sessionOnce invoked
Fable 5 $0.01981 $0.01981
Opus 5 $0.00991 $0.00991
Sonnet 5 $0.00396 $0.00396
Haiku 4.5 $0.00198 $0.00198

Measured yesterday against content hash 1e40b936d0a3, method: parsed. Prices are Anthropic first-party input rates as of 2026-08-30, from the pricing page.

Security

Grade A, and why

agent-controller AGENTS.md scanned grade A with 0 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured yesterday.

A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.

Nothing flagged

None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.

AGENTS.md · 137 lines

How it starts

The opening of the file, as written. The whole thing — 137 lines — stays where its author put it; the contents beside it link to each section on GitHub.

AGENTS.md

Guidance for AI agents (and human contributors) working in this repository. This is the canonical, tool-agnostic guide; CLAUDE.md imports it.

What this is

Agent Controller is a declarative runtime for AI agents: define an agent in YAML (ADL — Agent Definition Language) and run the same spec on the Pi, opencode, Codex, or Claude runtime through a consistent backend interface. The agentctl Go binary compiles a spec and dispatches it over a versioned stdio NDJSON wire protocol to a Node runtime adapter, which loads the local registry (tools, extensions, skills, agents, MCP servers) and drives the session against the model provider.

Repository layout

Path What
cli/ agentctl — Go binary (validate / compile / run / chat / serve).
runtime/ Pi runtime adapter (Node/TS). runtime.type: local.
runtime-opencode/ opencode adapter (Node/TS). runtime.type: local-opencode.
runtime-codex/ Codex adapter (Node/TS). runtime.type: local-codex (requires model.provider: openai).
runtime-claude/ Claude Agent SDK adapter (Node/TS). runtime.type: local-claude (requires model.provider: anthropic; no external CLI on PATH).
schemas/ ADL + manifest JSON Schemas — the source of truth.
examples/ Example ADL specs + scheduler integrations (Maestro / Airflow / Temporal).
skills/, tools/, extensions/, agents/ Local registry the adapters load.
e2e/ End-to-end harness (ADAPTER=pi|opencode|codex|claude).
docs/architecture/ Overview, wire protocol, per-adapter capability matrix.

Setup, build, test

Requires Go ≥ 1.25 and Node ≥ 22.19.0 + npm (the strictest adapter engines floor; runtime-opencode / runtime-codex / runtime-claude only need ≥ 22).

# Build
(cd runtime          && npm install --ignore-scripts && npm run build)
(cd runtime-opencode && npm install --ignore-scripts && npm run build)
(cd runtime-codex    && npm install --ignore-scripts && npm run build)
(cd runtime-claude   && npm install --ignore-scripts && npm run build)
(cd cli && go build -o bin/agentctl ./cmd/agentctl)

# Test
(cd runtime && npm test)            # vitest
(cd runtime-opencode && npm test)   # builds, then vitest
(cd runtime-codex && npm test)      # builds, then vitest
(cd runtime-claude && npm test)     # builds, then vitest
(cd cli && go test ./...)           # includes the schema-sync test

Read the full file on GitHub · 137 lines

Changes

What this file has done since we first saw it

Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.

  1. yesterday First seen · 137 lines · 1,981 tokens per session scan A 1e40b936d0a3

Subscribe to this mod's changes

agent-controller AGENTS.md is an instructions file published in the GitHub repository CCDevelopForFun/agent-controller (2 stars, last pushed 3d ago), licensed MIT. It adds 1,981 tokens to every session, about $0.0099 per session on Opus 5. A static security scan graded it A with 0 findings. No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-31.